A tamper detection method, apparatus, equipment, program product, and storage medium

By combining histogram equalization, CNN, SRM filter, and multiple edge detection operators with longitudinal attention and cross-attention mechanisms, the problem of low accuracy in electricity meter image tampering detection is solved, and efficient electricity meter tampering detection is achieved.

CN122090428APending Publication Date: 2026-05-26CHINA MOBILE COMM LTD RES INST +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
CHINA MOBILE COMM LTD RES INST
Filing Date
2024-11-26
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing image tampering detection algorithms lack understanding of image content in electricity meter images and fail to effectively utilize semantic information, resulting in low detection accuracy and difficulty in detecting subtle tampering of electricity meter images.

Method used

Histogram equalization is used to remove reflections, CNN is used to extract contextual features of bounding box images, SRM filters are used to extract noise features, multiple edge detection operators are used to extract edge features, and vertical attention and cross attention mechanisms are combined to determine whether the image has been tampered with.

Benefits of technology

It improves the accuracy of meter image tampering detection, effectively identifies tampered areas of meter numbers, saves labor costs, and reduces electricity bill verification costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122090428A_ABST
    Figure CN122090428A_ABST
Patent Text Reader

Abstract

This application discloses a tampering detection method, apparatus, device, program product, and storage medium. The method includes: acquiring a second image corresponding to a first image; the second image being a suspected tampering portion of the first image; acquiring a feature set of the second image, the feature set including context features, noise features, and edge features; determining vertical attention features and cross-attention features of the second image based on the feature set; and determining whether the first image has been tampered with based on the vertical attention features and the cross-attention features.
Need to check novelty before this filing date? Find Prior Art